The paper in Nature Ecology & Evolution is here: go.nature.com/2BhsuRA

Here was the problem: Mortality rates of forest tree species are biased towards the small individuals. If you examine the distribution of sizes of any forest, you will see that the vast majority of stems are small, and it is well established that smaller trees die more often which leads to the bias in mortality rates. The flip-side of this problem is that there are not that many big trees which lead to less opportunity to observe mortality. The big trees are where most of the biomass resides and thus carbon is stored, and having a deep understanding of the fate of carbon in the forest is crucial to forecasting the future of the carbon cycle. Now, set this problem in the vastly complex tropical forests where there are thousands of species with who knows how many life history strategies. We set out to provide a means of modelling survival of trees as a function of the size of the individual that would avoid the bias toward small tree mortality and provide forecasters more aggregated information about survival in tropical tree species. 

During a ForestGEO workshop in Hainan Island, China sponsored by the national science foundations of the United States and China, and the Smithsonian Institution, we set to work on overcoming the small tree bias and deepening our understanding of tropical tree survival. Leveraging new methods that address size biases in the data, we fit functions that estimate small and large tree survival in a way that can be very useful for predicting overall forest dynamics.

Our analysis emerged from a large collaborative effort made possible by the ForestGEO network [https://forestgeo.si.edu/]. Research networks are critical to this kind of study because they provide amazing amounts of data that are standardized. Every plot is implementing the same protocol, and contributing their data to the common cause of global analyses. Further, there are experts at every location that can help such a synthesis make sense of how forests share common patterns in survival, as well as how they might differ.  

To better understand similarities in tropical tree survival, we grouped species into clusters based on their survival function parameters. We also wanted to know if these clusters or as we call them ‘survival modes’ of species shared common traits, but did not find a strong relation between trait values and survival mode. Finally, we wanted to know if there were climate correlates that related the abundance of survival modes. We did find a relation between mean annual temperature and cumulative water deficit and the relative abundance of the Large Canopy species suggesting that this mode of survival is more successful in cooler regions of the tropics with longer dry seasons.

 (photo credit Charlie Koven)

Finally, we wanted to know if these observational data could help inform how earth system models represent vegetation in the tropics. We compared our observed results to a model designed to predict the dynamics of forests globally  (the Functionally-Assembled Terrestrial Ecosystem Simulator  [FATES], developed by researchers at the US Department of Energy and the National Center for Atmospheric Research). We discovered FATES under-predicted tropical forest mortality patterns for large diameter trees. Not only can our approach identify potential mismatches between predictions and observations, but offers pathways to improving these global models. By focusing on important modes of survival, and not just site or species averages, we are in a position to improve predictions of how tropical forests operate.

Moving forward, we hope to expand this approach to complex forests using demography into other life-history strategies, such as growth and reproduction. We know now, however, that this research must focus on the large and small ways that trees live life.